bmad-spec

Distill intent inputs into a five-field SPEC.md kernel with companion files and an append-only memlog.

Updated Jul 9, 2026
One-click install
npx skills add https://github.com/unbakedcosmos/thumbnailer --skill bmad-spec-unbakedcosmos
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: bmad-spec
Source: https://github.com/unbakedcosmos/thumbnailer/tree/main/.claude/skills/bmad-spec
Command: npx skills add https://github.com/unbakedcosmos/thumbnailer --skill bmad-spec-unbakedcosmos

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes assets (resource) components.

What problem does it solve? Turning vague ideas, brain dumps, PRDs, transcripts, or mixed multi-source material into a single canonical specification is error-prone: decisions get lost, capability IDs drift, and downstream planning steps consume inconsistent documents. This Skill produces a preservation-validated SPEC.md contract that every downstream BMad workflow can rely on. ## Core Features & Use Cases - Five-field kernel distillation: Converts any intent input into SPEC.md carrying Why, Capabilities (with stable CAP-N IDs), Constraints, Non-goals, and a testable Success signal. - Companion file management: Routes load-bearing overflow content (glossaries, diagrams, entity catalogs) into named companion files referenced from SPEC.md frontmatter. - Append-only memlog derivation: Records every decision in .memlog.md and re-derives SPEC.md from it, so updates never cause merge drift and capability IDs stay stable. - Two-pass self-validation: Checks coherence against eight Spec Law rules and walks the source claim-by-claim to confirm preservation. - Use Case: Paste a customer email thread and a meeting transcript, and receive a spec folder with SPEC.md, a glossary companion, and a memlog recording every decision, assumption, and open question. ## Quick Start Ask the assistant to distill the attached product brief into a BMad spec and validate it against Spec Law.

Frequently Asked Questions about bmad-spec

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I turn a vague product idea into a formal specification?▼

Provide the idea, brain dump, or source document as input and the skill distills it into a SPEC.md kernel with Why, Capabilities, Constraints, Non-goals, and a Success signal. Sparse inputs are handled in express mode, with gaps recorded as open questions.

What is the difference between SPEC.md and the memlog?▼

The .memlog.md file is the canonical append-only record of every decision, capability, and assumption in chronological order. SPEC.md is derived from the memlog on each run and is never hand-edited, which prevents merge drift across updates.

Can multiple skills update the same spec over time?▼

Yes. A second invocation with the same slug lands at the existing spec folder and updates it in place. Existing capability IDs are preserved, new capabilities get the next unused CAP-N, and retired IDs are never reused.

When should content go into a companion file instead of SPEC.md?▼

Content that needs more than one kernel-shaped line, such as multi-item catalogs, tables, diagrams, or glossaries, belongs in a companion file named for its content type. Diagrams always land in a companion regardless of size.

What happens if the input is too thin to distill?▼

The skill stops rather than inventing content. In headless mode it returns a blocked status with error_code insufficient_intent, and in interactive mode it suggests using bmad-prd to draw out the vision first.